Truth or .... False Color

[David Battistella]

So what we are really saying here is that the RED MX sensor in RAW view is a 320 ISO sensor (can be rated from 320-1000 or whatever you like), but none of the onboard exposure tools are relevant to the RAW view?

They only refer to the way you have set up your camera meta data for the way you want to expose it?

are we asking that red standardize their exposure tools?

if you are looking at this on a scope directly from the camera, then you are not seeing the RAW sensor data, your monitoring a feed with two different meta data settings applied (one raw view one with rg3/rg3 or however you have it set). That signal going into the scope is not what the sensor is seeing, it's a monitoring path, like a one light, it's not everything that is on the neg(in the raw file container).

You can't see that until you look at an r3d in RCX and start moving it around.

What you are asking for is quite tricky and I can completely see why it was set up this way in the first place.

Battistella

I would tell you that on MX sensor - RAW view is close to 320 - but mapped very differently in the highlights - just toggle between the two on a scope - VERY different. NOT the same.

It is an RGB representation that I strongly suspect was based on a method for matching M sensors and MX sensors. It AIN'T "RAW" - and ALL tools in this view are NOT based on 320.

 

[Mark L. Pederson]

However, without getting into details, there are some limitations and this will not always be perfect (especially when monitoring at extremely low/high ISOs, etc).

Trent - For the record, I see EXTREME accuracy at high and low ISOs when in RAW view mode - and I have tested those results extensively in post processing tools with perfect success - so ... whatever you are doing with the EXPOSURE false color when RAW view in enabled - is EXACTLY what we want even when not in RAW view.

 

[Jake Wilganowski]

The intention is that the red and purple bands in Exposure tool act on the RAW data. However, without getting into details, there are some limitations and this will not always be perfect (especially when monitoring at extremely low/high ISOs, etc). We will run through more tests here to see if the current behavior needs to be corrected.

It sounds like we have confirmed it changes with iso instead of sticking to raw. So, YES, it needs to be not only corrected, but announced NOW so that working DP's aren't getting screwed on set. And, I just have to ask, how the heck does something this major get released? I mean, it's one of a handful of extremely important exposure tools.

Sorry for ranting but I use this mode often and am thinking back to my last jobs. It really upsets me if I was seeing inaccurate info and making decisions on it for commercial work. Stop the presses and fix it please !!

 

[Mark L. Pederson]

So what we are really saying here is that the RED MX sensor in RAW view is a 320 ISO sensor (can be rated from 320-1000 or whatever you like), but none of the onboard exposure tools are relevant to the RAW view?

No.

David - we have a failure to communicate.

I'll try again - currently, when in RAW view - regardless of your ISO - your tools display values based as closely as they can to photons hitting the sensor or lack thereof. So ... it is a way to see your over/under very specifically. With much more detail than the stoplights - which will trigger with ANY amount of clip - even a small specular highlight. But we can only achieve this when in the RAW view. The RAW view on MX is NOT ISO 320. Just look at scope - 320 ISO vs RAW view. Not the same.

are we asking that red standardize their exposure tools?

No. What I am asking for is extremely logical. And RED is already saying it should work the way I want it to work. There's clearly a bug.

if you are looking at this on a scope directly from the camera, then you are not seeing the RAW sensor data, your monitoring a feed with two different meta data settings applied (one raw view one with rg3/rg3 or however you have it set). That signal going into the scope is not what the sensor is seeing, it's a monitoring path, like a one light, it's not everything that is on the neg(in the raw file container).

You can't see that until you look at an r3d in RCX and start moving it around.

What you are asking for is quite tricky
Battistella

David - tricky or not - what I am asking for already works - but it only works when you go to view RAW mode. The WHOLE POINT of this tool is, regardless of ISO settings and monitor path choice - the camera can ASSIST you in choosing your EXPOSURE by indicating when not enough light is hitting the photons vs. too much light is hitting the photons.

Stoplights ... over the whole image

 

[Hans von Sonntag]

Ultimately, my own personal methods are simple.

1. Leave it at 320 to 800 ISO depending on the scene circumstances.
2. Use RAW view and histogram and goal posts to judge exposure.
3. Use 1/8 blue filtration on all tungsten only and some mixed lighting scenes.
4. Treat it like film and know which scenes you can push/pull based on your noise threshold preference....


So what you are asking for is a bit more rigid system to make it easier and more accessible to more people?

Battistella

David, of course you are using the same methodology as I do, or Gunleik, Phil, and many others here on the boards. But outside, in the rooms of rental houses, on the sets of TVC land, in remote places of documentary land, and wherever, you see the most adventurously methodologies to shoot with Red cameras. People dial in 1200 ASA and then overexpose and thus achieve good exposure. They use RedLogFilm and underexpose, telling the clients that Red has to look badly washed out. I've seen footage with curves applied and footage that was noisy as hell because they used ISO 800 and managed somehow to underexpose. They said the pictures looked bright enough in the viewfinder. They used RedRaw. Others use curves, tint and whatnot and translate that into their dailies for later grading. There are million ways to screw up the footage often conducted by pro DITs that learned a lot of FUD spread from set to set.

Yes I'm asking for a more ridged system that can be opened up for the seasoned pro who really knows what she/he does. I have my Epic set to 400 ASA, 5600 K, RedColour3, RedGamma3 and use later in post RedlogFilm with mentioned settings. I never change those settings. RedlogFilm is hardly usable for shooting but meaningful for evaluating exposure in high contrast environments.

Hans

 

[Scott Crawley]

Having an exposure tool that reflects sensor/raw clip and noise/floor would reassure anyone.

Agreed. Isn't that what RAW view does? If the tools are not accurately representing the RAW data while in RAW view then I see the issue.

Please explain what method you use now to "reassure a client RE the exposure choices now"?

If they are monitoring Red Gamma/Red Color (99.999% of the time) and they are not entirely comfortable with our choices we flip over to RAW to more accurately assess the captured image before metadata alterations are applied.

What I am suggesting - the ability to show your client or not show them - the Exposure Tool over the image they are looking at [snip] - without having to disrupt anyone with the ugly ass RAW RGB representation.

I understand. My contention is that an anylitical tool must be empirically correct; an exact representation of the measured subject and not an interpretation of it. If you are monitoring a Red Gamma/Color image then the tools must represent that image, as displayed. Showing a RAW analysis over a Red Gamma/Red Color image is not correct (strictly speaking) and therefore is an inaccurate representation. It is a "false" representation, if you will. Do you find our points of view on this point to be at odds with one another, or do I misunderstand your point?

Maybe the name "False Color" is the root of the problem?

The RAW image is ugly. I try not to show it myself but it is only a problem, really a problem, if the client can't understand that it is an undeveloped image more akin to a negative than a positive film frame. Most people educated in photography get that. Those that don't simply don't understand RAW and shouldn't see it.

 

[Mark L. Pederson]

Agreed. Isn't that what RAW view does? If the tools are not accurately representing the RAW data while in RAW view then I see the issue.



If they are monitoring Red Gamma/Red Color (99.999% of the time) and they are not entirely comfortable with our choices we flip over to RAW to more accurately assess the captured image before metadata alterations are applied.



I understand. My contention is that an anylitical tool must be empirically correct; an exact representation of the measured subject and not an interpretation of it. If you are monitoring a Red Gamma/Color image then the tools must represent that image, as displayed. Showing a RAW analysis of a Red Gamma/Red Color representation is not correct (strictly speaking) and therefore is an inaccurate representation. It is a "false" representation, if you will. Do you find our points of view on this point to be at odds with one another, or do I misunderstand your point?

Maybe the name "False Color" is the root of the problem?

Scott - your contention is not my contention. The stoplights do not "represent that image, as displayed" do they?

Again - if we had the ability to get RAW view on the LCD with RC3 out the monitor path at the same time - I would not have bothered to start this thread. But on most productions - sending RAW view down the monitor path causes confusion and leads to mistakes. Many times AC's forget to switch back - and clients and DITs start freeking out and/or calling for adjustments, etc.

RED already has VIDEO - under TOOLS and that gives you "False Color" based on IRE/ISO.

You can't view "RAW" - so ... anyway you slice it - you must create an RGB representation based on some logic. I think RED's logic for their RAW RGB interpretation is based on matching M and MX - I could be wrong.

It would be far less confusing for everyone if they would have made the choice to display a LOG representation - NOT tied to IRE/ISO - and just called it LOG instead of RAW. And again - they are gonna have to do that ANYWAY if they want CDL and 3D LUT support aren't they

 

[Maz Mawlawi]

Trent - For the record, I see EXTREME accuracy at high and low ISOs when in RAW view mode - and I have tested those results extensively in post processing tools with perfect success - so ... whatever you are doing with the EXPOSURE false color when RAW view in enabled - is EXACTLY what we want even when not in RAW view.

Yep...this pretty much sums it up Trent ! It used to work that way and we relied on it all the time...

 

[Scott Crawley]

Scott - your contention is not my contention. The stoplights do not "represent that image, as displayed" do they?

Nope they don't. I agree the stop lights are incorrect in Red Gamma/Color and can be confusing to people.

Again - if we had the ability to get RAW view on the LCD with RC3 out the monitor path at the same time - I would not have bothered to start this thread.

Yes. I would love that too. :thumbsup:

You can't view "RAW" - so ... anyway you slice it - you must create an RGB representation based on some logic.

Yes of course, but that is another can of worms that we don't open outside of the camera dept. even with the technically savvy.

 

[Mark L. Pederson]

Nope they don't. I agree the stop lights are incorrect in Red Gamma/Color and can be confusing to people.

Not sure how these are confusing when the very, sole purpose is to indicate when the sensor itself is clipping in channels. I find these tools critically important and I have NEVER seen anyone confused by them.

There would be ZERO point to even having them if they were downstream from IRE/ISO processing.

 

[Scott Crawley]

I don't know anyone who is confused by them, but they CAN BE confusing if people think they are representative of the monitored image. These stop lights are fairly unique to Red and it is easy enough to understand that they are always indicative of the sensor. The same can not be said of histograms and color overlays. When folks look at those they automatically assume they are looking at an analysis of the monitor feed. At least that has been my experience.

At any rate, I don't disagree; the option to send RG3 out the monitor while keeping RAW to myself would be a very good thing. I just wouldn't otherwise change the way the analytical tools function so long as they are correct. FWIW

Not sure how these are confusing when the very, sole purpose is to indicate when the sensor itself is clipping in channels. I find these tools critically important and I have NEVER seen anyone confused by them.

There would be ZERO point to even having them if they were downstream from IRE/ISO processing.

 

[Mark L. Pederson]

I just wouldn't otherwise change the way the analytical tools function. FWIW

Would you not admit that there is ZERO negative about making the EXPOSURE false color based on sensor - like it is when in RAW mode - since you already have "VIDEO" which gives you false color based on IRE/ISO. You don't loose anything, right?

You don't gain anything from the RGB representation of RAW, do you?

 

[David Battistella]

Some conversations are better on the telephone.

I get everything you are saying. I'm all for the exposure tool working only on the RAW sensor data (and currently only on RAW view).

I guess what I am saying is this.

If I am in RAW view the only things that should affect the false color reading are,
1. Change in the aperture
2. Change the shutter angle.

If I change the ISO, it should not change, because it is not affecting the photo's hitting the sensor.

So you just want false color to work base on RAW view without actually switching to RAW view because currently, false color displays based on meta settings? (BAD, I agree)

on the ISO320 thing, what ISO do you think RAW view is showing us? It's changed over the years for sure.

Battistella

No.

David - we have a failure to communicate.

I'll try again - currently, when in RAW view - regardless of your ISO - your tools display values based as closely as they can to photons hitting the sensor or lack thereof. So ... it is a way to see your over/under very specifically. With much more detail than the stoplights - which will trigger with ANY amount of clip - even a small specular highlight. But we can only achieve this when in the RAW view. The RAW view on MX is NOT ISO 320. Just look at scope - 320 ISO vs RAW view. Not the same.

Stoplights ... over the whole image

 

[Mark L. Pederson]

So you just want false color to work base on RAW view without actually switching to RAW view because currently, false color displays based on meta settings? (BAD, I agree)

on the ISO320 thing, what ISO do you think RAW view is showing us? It's changed over the years for sure.

Battistella

Didn't I say that in my first post

RAW view is close to ISO320 - but highlights specifically are mapped a bit higher than RG3 at ISO 320. I don't have the patience to reverse engineer what I think should just be eliminated

 

[paulherrin]

forget iso... iso is imaginary, i don't think it applies at all. my understanding is that the raw view is simply a map of what the sensor is collecting... no?

as per the histogram/etc, we don't want to forget about the guys throwing out to an external recorder only, but i think the video mode handles that...

 

[Scott Crawley]

Not really. I think it would ultimately add to the confusion. If my Director is looking at a feed that is clearly blown out, but the overlay indicates that it is not, then he may choose not to trust the tool at all. I could be wrong but I think it is plausible and even likely.

I would sooner suggest just eliminating false color over RG3 all together as the most simple solution. IMO RAW view is the only place to make exposure descisions (edit: On a RAW only camera), so exposure tools really have no place anywhere else. Would you agree?

People will still try to make exposure decisions based upon the RG3 feed using the histogram and distinctions will have to be made, but having that sensor view is enough to set things straight. I see the argument for renaming RAW view to something like "sensor view" in that case if folks think that would really help.

Would you not admit that there is ZERO negative about making the EXPOSURE false color based on sensor - like it is when in RAW mode - since you already have "VIDEO" which gives you false color based on IRE/ISO. You don't loose anything, right?

You don't gain anything from the RGB representation of RAW, do you?

 

[Scott Crawley]

I wouldn't call ISO imaginary, but I believe Graeme has explained why it doesn't really apply to digital photography. ISO was only retained in attempt to keep things familiar. It will fade away. I think that is inevitable.

forget iso... iso is imaginary, i don't think it applies at all. my understanding is that the raw view is simply a map of what the sensor is collecting... no?

as per the histogram/etc, we don't want to forget about the guys throwing out to an external recorder only, but i think the video mode handles that...

I was forgetting about external recorders. It might be another argument for keeping the analytical tools tied to the monitored signal. I am ignorant with regard to a Video Mode because I never use it. If you say it addresses the need then I'll take your word for it.

 

[Mark L. Pederson]

People will still try to make exposure decisions based upon the RG3 feed using the histogram and distinctions will have to be made, but having that sensor view is enough to set things straight. I see the argument for renaming RAW view to something like "sensor view" in that case if folks think that would really help.

RED should kill two birds with one stone.

The very FIRST STEP for RED to enable CDL and 3D LUT support that RED very much needs to implement - is a monitor signal that is LOG encoded and is not effected by any meta-data settings (except ISO + White Balance) - kinda like a locked off version of RedLOG film. Kinda like LOG-C

Call it "LOG" - which would be correct, and free of confusion. Eliminate RAW view - as clearly - just based on posts on this thread - created confusion. And nobody can tell me why they need or want the RGB representation other than to get accurate tools that should be available over RG3, etc. if wanted.

Then -

Then, you can have it my way - EXPOSURE works over any monitor signal - tied to sensor - free of ISO/meta.

And ... in your case, you could toggle to LOG with the EXPOSURE overlay.

Or ... VIDEO false color which is all IRE/ISO/meta effected.

In my word - we don't want the overlay (EXPOSURE false color) on the client monitors EVER - we only want it for the camera department - and we don't want to switch the clients monitor path to get to get it.

 

[Scott Crawley]

RED should kill two birds with one stone.

The very FIRST STEP for RED to enable CDL and 3D LUT support that RED very much needs to implement - is a monitor signal that is LOG encoded and is not effected by any meta-data settings (except ISO) - kinda like a locked off version of RedLOG film.

Call it "LOG" - which would be correct, and free of confusion. Eliminate RAW view - as clearly - just based on posts on this thread - created confusion. And nobody can tell me why they need or want the RGB representation other than to get accurate tools that should be available over RG3, etc. if wanted.

I can see much wisdom in that. FWIW

Still 'togs have pretty much all gotten used to the term RAW and what that entails. Can't the motion picture community do the same?

 

[Mark L. Pederson]

Still 'togs have pretty much all gotten used to the term RAW and what that entails.

I dunno ... I haven't heard one correct explanation of the RGB representation or logic used to create that representation for the RAW view on RED cameras by anyone. I'm not saying that the logic used is flawed - just saying nobody understands it other than "you can't view RAW so you need to do something". All people really know is that RAW view is where the exposure tools do work independent from ISO/Meta. Many folks currently, incorrectly think that is actually the case outside of RAW view.